Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Add demo script for single-node async checkpointing #200

Merged
merged 8 commits into from
Nov 22, 2024

Conversation

awonak
Copy link
Contributor

@awonak awonak commented Nov 20, 2024

Fixes b/369874146

Example demo outputs:

$ python demo/checkpointing/train.py --project=gcs-tess --bucket=awonak-us-west1 --embedding-dim=10 --num-epochs=3 
Epoch 1/3, Train Loss: 449945.0625
Epoch 2/3, Train Loss: 409629.6875
Epoch 3/3, Train Loss: 380471.9375
Average checkpoint save time: 0.8570 seconds (stdev 0.1997)
Total run time: 1m53s

$ python demo/checkpointing/train.py --project=gcs-tess --bucket=awonak-us-west1 --embedding-dim=10 --num-epochs=3 --use-async
Epoch 1/3, Train Loss: 456512.0000
Epoch 2/3, Train Loss: 413212.9375
Epoch 3/3, Train Loss: 382256.8125
Average checkpoint save time: 0.0009 seconds (stdev 0.0009)
Total run time: 1m36s

Note: I will update the main README with single node async details once either this or #190 gets merged.

  • Tests pass
  • Appropriate changes to documentation are included in the PR

@awonak awonak marked this pull request as ready for review November 20, 2024 23:16
@awonak awonak requested a review from a team as a code owner November 20, 2024 23:16
@awonak awonak requested review from Yash9060 and jdnurme November 20, 2024 23:16
@awonak awonak merged commit 336fedf into main Nov 22, 2024
5 checks passed
@awonak awonak deleted the awonak-singlenode-async branch November 22, 2024 18:07
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

3 participants